A measurement-system analysis asks whether the observed measurement differences are small enough relative to the part-to-part differences and the intended decision. The most familiar output is Gage R&R, but the result becomes useful only when its components are understood.
The crossed study structure
In a typical crossed Gage R&R study, every operator measures every selected part multiple times. The study needs parts that represent the process range, operators who normally use the method, repeated trials, and a measurement order that reduces memory and sequence effects.
Repeatability
Repeatability is the variation observed when the same operator measures the same part repeatedly with the same measurement system under the study conditions. High repeatability variation may point toward equipment resolution, fixture stability, part positioning, contact method, environmental sensitivity, or an unclear measurement technique.
Reproducibility
Reproducibility is the variation associated with differences between operators or appraisers. It can include systematic operator offsets and, depending on the method, operator-by-part interaction. High reproducibility variation may point toward inconsistent technique, datum interpretation, force, alignment, reading method, or insufficient work instructions.
What %GRR is being compared with?
Reports may show Gage R&R as a percentage of study variation, a percentage of tolerance, or both. These percentages answer related but different questions.
| Metric | Question |
|---|---|
| %Study Variation | How large is measurement variation relative to the total variation observed in the study? |
| %Tolerance | How much of the engineering tolerance is consumed by measurement variation? |
A study with a narrow part range can make the measurement system look large relative to study variation even when its tolerance contribution is smaller. That does not automatically make the study acceptable; it may mean the selected parts did not represent the needed process range.
Number of distinct categories
ndc estimates how many categories of part variation the measurement system can distinguish. A low ndc means the system has limited discrimination relative to the part variation in the study. The value should be read alongside the part selection and Gage R&R components.
Acceptance criteria need context
Common threshold bands are often taught, but the final decision depends on risk, tolerance, characteristic importance, process control needs, cost, and customer requirements. A measurement system used for process improvement may need stronger discrimination than one used only for a broad screening decision.
When the result is poor
- Check for data-entry, part-ID, operator-ID, or trial-order errors.
- Review equipment resolution, calibration status, and fixture condition.
- Observe the measurement method directly.
- Compare operator means and ranges.
- Look for part features that interact with operator technique.
- Improve the method, then repeat the study with representative parts.
What a useful report should state
Include the characteristic, unit, tolerance, equipment, resolution, parts, operators, trials, study type, calculation method, results, plots, investigation notes, and conclusion. A red/green status without this context is difficult to audit or improve.
Do not improve the percentage by narrowing the study artificially. The study should represent the measurement decision and process range that matter in real use.